mirror/eigen
2
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-03-13 18:37:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

ARM NEON supports multiply-accumulate instruction vmla, use that in pmadd().

Browse Source
This commit is contained in:
kmargar 2012-05-28 14:55:23 +03:00
parent b202c5ed2f
commit 97cdf6ce9e
2 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
CMakeLists.txt
View File

@ -164,7 +164,7 @@ if(CMAKE_COMPILER_IS_GNUCXX)
option(EIGEN_TEST_NEON "Enable/Disable Neon in tests/examples" OFF)
if(EIGEN_TEST_NEON)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mfloat-abi=softfp -mfpu=neon -mcpu=cortex-a8")
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mfpu=neon -mcpu=cortex-a8")
message(STATUS "Enabling NEON in tests/examples")
endif()

3
Eigen/src/Core/arch/NEON/PacketMath.h
View File

@ -160,7 +160,8 @@ template<> EIGEN_STRONG_INLINE Packet4i pdiv<Packet4i>(const Packet4i& /*a*/, co
}
// for some weird raisons, it has to be overloaded for packet of integers
template<> EIGEN_STRONG_INLINE Packet4i pmadd(const Packet4i& a, const Packet4i& b, const Packet4i& c) { return padd(pmul(a,b), c); }
template<> EIGEN_STRONG_INLINE Packet4f pmadd(const Packet4f& a, const Packet4f& b, const Packet4f& c) { return vmlaq_f32(c,a,b); }
template<> EIGEN_STRONG_INLINE Packet4i pmadd(const Packet4i& a, const Packet4i& b, const Packet4i& c) { return vmlaq_s32(c,a,b); }
template<> EIGEN_STRONG_INLINE Packet4f pmin<Packet4f>(const Packet4f& a, const Packet4f& b) { return vminq_f32(a,b); }
template<> EIGEN_STRONG_INLINE Packet4i pmin<Packet4i>(const Packet4i& a, const Packet4i& b) { return vminq_s32(a,b); }